Ontological Argument against Architecture without Implementation
Over the years, I've lived a double life. On one side, I've been a member of many "pure architecture" teams and forums. We couldn't see code with a pair of binoculars. On the other side, I've been knee-deep in the code trenches where emergent architecture changed and morphed daily. Architecture? What Architecture? All we had was code. That was all that mattered. Truly, the right path lies somewhere in between. However, when budget and time-to-market matter -- I would err on the side of functionality over architecture. Let the architectural decisions emerge from the the implementation. Now for the argument...
Many people are familiar with the ontological argument for God. This follows the same reasoning, proving once and for all that the best architecture must include implementation. For ease of translation, let's call the Optimal Architecture: Optimus.
Now, Optimus is an architecture of which no greater can be conceived.
As St. Anselm reasoned, if such an architecture fails to exist, then a greater architecture — namely, an architecture than which no greater can be conceived, and which exists — can be conceived. Therefore, Optimus must exist.
Let's now extend this argument and define terms. We can all agree that an architecture that exists in both diagrams and implementation is greater than an architecture that exists only in diagrams. Thus, Optimus existence must include implementation.
Therfore, all Optimal Architectures must exist in both diagrams as well as implementation.
Later on, I'll prove that Model-Driven-Architecture in fact proves the non-existence of Architects. See Hitchhiker's Guide for a preview.
(Yes, this is tongue and cheek -- Flamers, please flame off)